Flow assurance

Predicting Severe Slugging in Toe-Down Horizontal Wells

This paper presents the results of an experimental and modeling study that demonstrates the existence of severe slugging in systems with upward-inclined lateral flow paths, such as a toe-down well.


Severe slugging is an important flow-assurance issue, typically observed in offshore pipeline-riser systems. The consequences of severe slugging include flooding of downstream production facilities and an overall decrease in productivity. Severe slugging had been thought to be limited to systems with a downward-inclined pipeline and vertical, catenary, or lazy-S-shaped riser. This paper presents the results of an experimental and modeling study that demonstrates the existence of severe slugging in systems with upward-inclined lateral flow paths, such as a toe-down well.


Severe slugging in offshore pipeline-riser systems with downward-inclined pipelines has been intensely studied because of its serious consequences. Most such studies are related to pipeline-riser systems, but the possibility of severe slugging in toe-up horizontal wells also was demonstrated experimentally in the literature. Toe-up wells are geometrically analogous to pipeline-riser systems because the lateral section in a toe-up well is a mostly downward-inclined flow path. However, there has been no documented evidence of severe slugging in systems with upward-inclined flow paths.

Facility Description

A large-scale experimental facility was used in this study. This facility was designed primarily to study flow behavior in horizontal wells. The lateral section in the facility was 236 ft long and comprised 6-in. inner-diameter (ID) pipe, which acted as the casing. The curvature section was made of bent acrylic pipe and had a radius of curvature of approximately 18 ft. The vertical section was 34 ft high with 2-in. ID polycarbonate pipe simulating the tubing. The schematic shows a packerless configuration. In such a configuration, an additional volume was necessary to simulate the effect of an annular volume. This was achieved by adding a 37-ft-high, 6-in.-ID expansion volume. For cases where a packer was required, a solid plug was placed in the system to simulate the packer and the annular volume was disconnected. The end of the tubing location was varied by pulling the tubing or running the tubing deeper.

The test fluids were air and water. Air and water entered the facility at the toe of the well. The mass flow rates were measured using flowmeters and were held constant during the tests. As the flow pattern developed along the lateral section, flowing pressures, temperatures, and pressure gradient were measured. Conductivity probes allowed the measurement of slug characteristics. The test facility also included several quick-closing valves. These valves were pneumatically operated and could capture the flow and measure holdup. When the flow reached the top of the vertical section, it returned to the water tank where the air was vented. A separate metering skid was used to inject gas in the tubing to study the effect of gas lift.

Test Matrix

Three end-of-tubing (EOT) locations were studied:

  • EOT 1: 39 ft inside the lateral section
  • EOT 2: end of curvature
  • EOT 3: bottom of the vertical section

Packered and packerless configurations were used to determine the effect of a packer. When the EOT was placed in the lateral, an additional test was performed, with the tubing centralized in the casing without the use of a packer, to study the effect of tubing centralization.

Experimental Results

Severe slugging was first observed in EOT 1 without a packer. At very low gas rates, severe slugging is observed, with the peak of the pressure fluctuations approaching the pressure observed during single-phase liquid flow. Severe slugging only occurs at the lowest gas velocity in this case. The chances of severe slugging are higher at higher liquid velocities. No severe slugging was observed for this EOT configuration at the lowest superficial liquid velocity (0.1 ft/sec).

Important flow characteristics were observed during the severe slugging process. The EOT is submerged during the buildup stage of severe slugging. As a result, the gas flows into the annulus while the liquid fills up the tubing. Once liquid production begins, gas enters into the tubing and, as a result, the EOT submergence is reduced. In cases where there was no severe slugging, the tubing was never completely submerged in liquid in the lateral (Fig. 1).

Fig. 1—No EOT submergence (EOT 1, packerless).


When the EOT was moved to shallower locations, as with EOT 2 and EOT 3, the chances of severe slugging were reduced. At higher gas velocities, lower liquid velocities, and shallower EOT locations, chances of severe slugging are reduced.


A packer in place was observed to eliminate severe slugging in all the experimental tests. The packer forces both liquid and gas into the tubing. Gas-lift injection eliminated severe slugging completely by reducing the buildup of liquid in the tubing.

The degree of submergence of the EOT is important in the severe-slugging process. To quantify the effect of submergence of the EOT experimentally, a separate test was conducted at EOT 1. The tubing was centralized using a mechanical fitting without the use of the packer. Therefore, the configuration was still packerless.

Centralization affects severe-slugging characteristics, demonstrating that the EOT flow dynamics also affect severe-slugging characteristics.


Two modeling efforts were undertaken as part of this study.

Transient Model. To predict the pressure behavior during severe slugging, a transient model is proposed. The model was developed for pipeline-riser systems with catenary risers. This model was chosen as the base model for modification because the geometry of a toe-down well closely resembles that of a pipeline-catenary-riser system, with the exception of the inclination of the lateral section. The following modifications were carried out to the model.

  • Modify mass-balance equations in the lateral to account for slug flow.
  • Remove the switch condition in the original model that accounts for liquid penetration back into the pipeline during severe slugging. This phenomenon was not observed physically in this study.
  • Account for the change of diameter along the flow path.
  • Develop a closure relationship to calculate liquid velocity in tubing.

The equations pertaining to the use of this model are detailed in the complete paper. The equations are discretized using an implicit forward difference numerical scheme. A moving grid method is adopted for special discretization. As a part of this method, the liquid column in the tubing is split into several nodes. The first node is fixed at the base of the riser, and the other nodes move up as the liquid column rises in the tubing.
The first set of calculations determines the steady-state variables, namely pressure, mixture velocity, and holdup. A marching algorithm is run from the top to the bottom of the liquid column to calculate steady-state values of the variables.

Upon calculation of steady-state variables, the iterative calculations for subsequent timesteps commence. Once the values from each timestep are calculated, the calculated values are used as predictor values for the next timestep. The timestep is then calculated for each iteration. Methodology for timestep calculation is provided in the complete paper.

Model Validation. The ­generated model was validated against the experimental data. The time period and the average amplitude of a pressure cycle recorded during the experiment were approximately 170 seconds and 13.4 psi, respectively. The model predicted a period of 234.4 seconds and an amplitude of 17.7 psi. A commercial multi­phase flow simulator predicted a period of 270 seconds and an amplitude of 14.96 psi.

Both the model and the commercial simulator do a fair job of predicting the pressure amplitude, but time-period predictions are better predicted by the simulator. With further experimentation and refinement of the closure relationship, the model predictions can be improved.

Single-Point Model. While a transient-pressure-prediction model is useful to predict amplitude and pressure behavior, a simple prediction tool is also useful. A criterion described in Eq. 1 of the complete paper is such a tool used for severe-slugging prediction in pipeline-riser systems. This criterion was modified to create a prediction tool for severe slugging in toe-down horizontal wells. The modifications include the following:

  1. Slug flow in the lateral
  2. Two-diameter (tubing/casing) configuration
  3. Reduced liquid velocity in tubing


  • In the experimental setup, severe slugging was observed only in packerless configurations. In the field, however, it is possible that severe slugging can occur in a packered system if the gas volume is large enough.
  • The submergence at the EOT was observed to be important. At very low gas rates, the film thickness in the slugs in the lateral is very high. If the tubing diameter is smaller than this thickness, it may result in the tubing being completely submerged. When the tubing is completely submerged, the liquid buildup stage will ensue.
  • An increase in superficial gas velocity and a reduction in the superficial liquid velocity reduces the chances of severe slugging.
  • The eccentricity of the EOT was also found to affect the severe-slugging pressure cycles.
  • Gas lift was observed to eliminate severe slugging.
  • Pulling the EOT to shallower depths will reduce the chances of severe slugging.
  • Transient and single-point models were developed to predict severe-slugging behavior in the system.
  • Both models predict the phenomenon acceptably, given the limited number of experimental data points for verification.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 191611, “Existence and Prediction of Severe Slugging in Toe-Down Horizontal Wells,” by Jayanth Nair, SPE, Eduardo Pereyra, SPE, and Cem Sarica, SPE, The University of Tulsa, prepared for the 2018 SPE Annual Technical Conference and Exhibition, Dallas, 24–26 September. The paper has not been peer reviewed.